Datasets:
Tasks:
Multiple Choice
Modalities:
Text
Formats:
parquet
Sub-tasks:
multiple-choice-qa
Size:
1K - 10K
License:
# coding=utf-8 | |
# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors. | |
# | |
# Licensed under the Apache License, Version 2.0 (the "License"); | |
# you may not use this file except in compliance with the License. | |
# You may obtain a copy of the License at | |
# | |
# http://www.apache.org/licenses/LICENSE-2.0 | |
# | |
# Unless required by applicable law or agreed to in writing, software | |
# distributed under the License is distributed on an "AS IS" BASIS, | |
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | |
# See the License for the specific language governing permissions and | |
# limitations under the License. | |
# Lint as: python3 | |
"""QA4MRE (CLEF 2011/2012/2013): a reading comprehension dataset.""" | |
import xml.etree.ElementTree as ET | |
import datasets | |
logger = datasets.logging.get_logger(__name__) | |
# pylint: disable=anomalous-backslash-in-string | |
_CITATION = r""" | |
@InProceedings{10.1007/978-3-642-40802-1_29, | |
author={Pe{\~{n}}as, Anselmoband Hovy, Eduardband Forner, Pamela and Rodrigo, {\'A}lvaro and Sutcliffe, Richard | |
and Morante, Roser}, | |
editor={Forner, Pamela and M{\"u}ller, Henning and Paredes, Roberto and Rosso, Paolo | |
and Stein, Benno}, | |
title={QA4MRE 2011-2013: Overview of Question Answering for Machine Reading Evaluation}, | |
booktitle={Information Access Evaluation. Multilinguality, Multimodality, and Visualization}, | |
year={2013}, | |
publisher={Springer Berlin Heidelberg}, | |
address={Berlin, Heidelberg}, | |
pages={303--320}, | |
abstract={This paper describes the methodology for testing the performance of Machine Reading systems through Question Answering and Reading Comprehension Tests. This was the attempt of the QA4MRE challenge which was run as a Lab at CLEF 2011--2013. The traditional QA task was replaced by a new Machine Reading task, whose intention was to ask questions that required a deep knowledge of individual short texts and in which systems were required to choose one answer, by analysing the corresponding test document in conjunction with background text collections provided by the organization. Four different tasks have been organized during these years: Main Task, Processing Modality and Negation for Machine Reading, Machine Reading of Biomedical Texts about Alzheimer's disease, and Entrance Exams. This paper describes their motivation, their goals, their methodology for preparing the data sets, their background collections, their metrics used for the evaluation, and the lessons learned along these three years.}, | |
isbn={978-3-642-40802-1} | |
} | |
""" | |
_DESCRIPTION = """ | |
QA4MRE dataset was created for the CLEF 2011/2012/2013 shared tasks to promote research in | |
question answering and reading comprehension. The dataset contains a supporting | |
passage and a set of questions corresponding to the passage. Multiple options | |
for answers are provided for each question, of which only one is correct. The | |
training and test datasets are available for the main track. | |
Additional gold standard documents are available for two pilot studies: one on | |
alzheimers data, and the other on entrance exams data. | |
""" | |
_BASE_URL = "http://nlp.uned.es/clef-qa/repository/js/scripts/downloadFile.php?file=/var/www/html/nlp/clef-qa/repository/resources/QA4MRE/" | |
PATHS = { | |
"2011": { | |
"_TRACKS": ("main"), | |
"_PATH_TMPL_MAIN_GS": "2011/Training_Data/Goldstandard/QA4MRE-2011-{}_GS.xml", | |
"_LANGUAGES_MAIN": ("DE", "EN", "ES", "IT", "RO"), | |
}, | |
"2012": { | |
"_TRACKS": ("main", "alzheimers"), | |
"_PATH_TMPL_MAIN_GS": "2012/Main_Task/Training_Data/Goldstandard/Used_in_Evaluation/QA4MRE-2012-{}_GS.xml", | |
"_LANGUAGES_MAIN": ("AR", "BG", "DE", "EN", "ES", "IT", "RO"), | |
"_PATH_ALZHEIMER": "2012/Pilot_Tasks/Biomedical_About_Alzheimer/Training_Data/Goldstandard/QA4MRE-2012_BIOMEDICAL_GS.xml", | |
}, | |
"2013": { | |
"_TRACKS": ("main", "alzheimers", "entrance_exam"), | |
"_PATH_TMPL_MAIN_GS": "2013/Main_Task/Training_Data/Goldstandard/QA4MRE-2013-{}_GS.xml", | |
"_LANGUAGES_MAIN": ("AR", "BG", "EN", "ES", "RO"), | |
"_PATH_ALZHEIMER": "2013/Biomedical_About_Alzheimer/Training_Data/Goldstandard/QA4MRE-2013_BIO_GS-RUN.xml", | |
"_PATH_ENTRANCE_EXAM": "2013/Entrance_Exams/Training_Data/Goldstandard/qa4mre-exam-test-withanswer.xml", | |
}, | |
} | |
def _get_question(topic_id, topic_name, test_id, document_id, document_str, question): | |
"""Gets instance ID and features for every question. | |
Args: | |
topic_id: string | |
topic_name: string | |
test_id: string | |
document_id: string | |
document_str: string | |
question: XML element for question | |
Returns: | |
id_: string. Unique ID for instance. | |
feats: dict of instance features | |
""" | |
question_id = question.attrib["q_id"] | |
for q_text in question.iter("q_str"): | |
question_str = q_text.text | |
possible_answers = list() | |
for answer in question.iter("answer"): | |
answer_id = answer.attrib["a_id"] | |
answer_str = answer.text | |
possible_answers.append({"answer_id": answer_id, "answer_str": answer_str}) | |
if "correct" in answer.attrib: | |
correct_answer_id = answer_id | |
correct_answer_str = answer_str | |
id_ = "_".join([topic_id, topic_name, test_id, question_id]) | |
logger.info("ID: %s", id_) | |
feats = { | |
"topic_id": topic_id, | |
"topic_name": topic_name, | |
"test_id": test_id, | |
"document_id": document_id, | |
"document_str": document_str, | |
"question_id": question_id, | |
"question_str": question_str, | |
"answer_options": possible_answers, | |
"correct_answer_id": correct_answer_id, | |
"correct_answer_str": correct_answer_str, | |
} | |
return id_, feats | |
class Qa4mreConfig(datasets.BuilderConfig): | |
"""BuilderConfig for Qa4mre.""" | |
def __init__(self, year, track="main", language="EN", **kwargs): | |
"""BuilderConfig for Qa4Mre. | |
Args: | |
year: string, year of dataset | |
track: string, the task track from PATHS[year]['_TRACKS']. | |
language: string, Acronym for language in the main task. | |
**kwargs: keyword arguments forwarded to super. | |
""" | |
if track.lower() not in PATHS[year]["_TRACKS"]: | |
raise ValueError("Incorrect track. Track should be one of the following: ", PATHS[year]["_TRACKS"]) | |
if track.lower() != "main" and language.upper() != "EN": | |
logger.warning("Only English documents available for pilot " "tracks. Setting English by default.") | |
language = "EN" | |
if track.lower() == "main" and language.upper() not in PATHS[year]["_LANGUAGES_MAIN"]: | |
raise ValueError( | |
"Incorrect language for the main track. Correct options: ", PATHS[year]["_LANGUAGES_MAIN"] | |
) | |
self.year = year | |
self.track = track.lower() | |
self.lang = language.upper() | |
name = self.year + "." + self.track + "." + self.lang | |
description = _DESCRIPTION | |
description += ("This configuration includes the {} track for {} language " "in {} year.").format( | |
self.track, self.lang, self.year | |
) | |
super(Qa4mreConfig, self).__init__( | |
name=name, description=description, version=datasets.Version("0.1.0"), **kwargs | |
) | |
class Qa4mre(datasets.GeneratorBasedBuilder): | |
"""QA4MRE dataset from CLEF shared tasks 2011, 2012, 2013.""" | |
BUILDER_CONFIGS = [ | |
Qa4mreConfig(year="2011", track="main", language="DE"), # 2011 Main track German (2011.main.DE) | |
Qa4mreConfig(year="2011", track="main", language="EN"), # 2011 Main track English (2011.main.EN) | |
Qa4mreConfig(year="2011", track="main", language="ES"), # 2011 Main track Spanish (2011.main.ES) | |
Qa4mreConfig(year="2011", track="main", language="IT"), # 2011 Main track Italian (2011.main.IT) | |
Qa4mreConfig(year="2011", track="main", language="RO"), # 2011 Main track Romanian (2011.main.RO) | |
Qa4mreConfig(year="2012", track="main", language="AR"), # 2012 Main track Arabic (2012.main.AR) | |
Qa4mreConfig(year="2012", track="main", language="BG"), # 2012 Main track Bulgarian (2012.main.BG) | |
Qa4mreConfig(year="2012", track="main", language="DE"), # 2012 Main track German (2012.main.DE) | |
Qa4mreConfig(year="2012", track="main", language="EN"), # 2012 Main track English (2012.main.EN) | |
Qa4mreConfig(year="2012", track="main", language="ES"), # 2012 Main track Spanish (2012.main.ES) | |
Qa4mreConfig(year="2012", track="main", language="IT"), # 2012 Main track Italian (2012.main.IT) | |
Qa4mreConfig(year="2012", track="main", language="RO"), # 2012 Main track Romanian (2012.main.RO) | |
Qa4mreConfig(year="2012", track="alzheimers", language="EN"), # (2012.alzheimers.EN) | |
Qa4mreConfig(year="2013", track="main", language="AR"), # 2013 Main track Arabic (2013.main.AR) | |
Qa4mreConfig(year="2013", track="main", language="BG"), # 2013 Main track Bulgarian (2013.main.BG) | |
Qa4mreConfig(year="2013", track="main", language="EN"), # 2013 Main track English (2013.main.EN) | |
Qa4mreConfig(year="2013", track="main", language="ES"), # 2013 Main track Spanish (2013.main.ES) | |
Qa4mreConfig(year="2013", track="main", language="RO"), # 2013 Main track Romanian (2013.main.RO) | |
Qa4mreConfig(year="2013", track="alzheimers", language="EN"), # (2013.alzheimers.EN) | |
Qa4mreConfig(year="2013", track="entrance_exam", language="EN"), # (2013.entrance_exam.EN) | |
] | |
def _info(self): | |
return datasets.DatasetInfo( | |
# This is the description that will appear on the datasets page. | |
description=_DESCRIPTION, | |
# datasets.features.FeatureConnectors | |
features=datasets.Features( | |
{ | |
"topic_id": datasets.Value("string"), | |
"topic_name": datasets.Value("string"), | |
"test_id": datasets.Value("string"), | |
"document_id": datasets.Value("string"), | |
"document_str": datasets.Value("string"), | |
"question_id": datasets.Value("string"), | |
"question_str": datasets.Value("string"), | |
"answer_options": datasets.features.Sequence( | |
{"answer_id": datasets.Value("string"), "answer_str": datasets.Value("string")} | |
), | |
"correct_answer_id": datasets.Value("string"), | |
"correct_answer_str": datasets.Value("string"), | |
} | |
), | |
# No default supervised keys because both passage and question are used | |
# to determine the correct answer. | |
supervised_keys=None, | |
homepage="http://nlp.uned.es/clef-qa/repository/pastCampaigns.php", | |
citation=_CITATION, | |
) | |
def _split_generators(self, dl_manager): | |
"""Returns SplitGenerators.""" | |
cfg = self.config | |
download_urls = dict() | |
if cfg.track == "main": | |
download_urls["{}.main.{}".format(cfg.year, cfg.lang)] = _BASE_URL + PATHS[cfg.year][ | |
"_PATH_TMPL_MAIN_GS" | |
].format(cfg.lang) | |
if cfg.year in ["2012", "2013"] and cfg.track == "alzheimers": | |
download_urls["{}.alzheimers.EN".format(cfg.year)] = _BASE_URL + PATHS[cfg.year]["_PATH_ALZHEIMER"] | |
if cfg.year == "2013" and cfg.track == "entrance_exam": | |
download_urls["2013.entrance_exam.EN"] = _BASE_URL + PATHS[cfg.year]["_PATH_ENTRANCE_EXAM"] | |
downloaded_files = dl_manager.download_and_extract(download_urls) | |
return [ | |
datasets.SplitGenerator( | |
name=datasets.Split.TRAIN, | |
gen_kwargs={"filepath": downloaded_files["{}.{}.{}".format(cfg.year, cfg.track, cfg.lang)]}, | |
) | |
] | |
def _generate_examples(self, filepath): | |
"""Yields examples.""" | |
with open(filepath, "rb") as f: | |
tree = ET.parse(f) | |
root = tree.getroot() # test-set | |
for topic in root: | |
topic_id = topic.attrib["t_id"] | |
topic_name = topic.attrib["t_name"] | |
for test in topic: | |
test_id = test.attrib["r_id"] | |
for document in test.iter("doc"): | |
document_id = document.attrib["d_id"] | |
document_str = document.text | |
for question in test.iter("q"): | |
yield _get_question(topic_id, topic_name, test_id, document_id, document_str, question) | |